Statistical Arbitrage Strategy with Cointegration

Overview

This is a simple Python program to understand the basics of Statistical Arbitrage (Stat Arb). It walks through:

1. Cointegration Test: Testing if two assets are cointegrated, which means their prices move together in the long term, making them suitable for a pairs trading strategy.
2. Modeling the Spread: Calculating the spread between two assets using linear regression.
3. Entry and Exit Signals: Generating trading signals when the spread deviates significantly from its mean.
4. Backtesting: Simulating the performance of the strategy over time and visualizing the portfolio performance.
5. Visualization: Plotting the portfolio value and the spread with entry and exit points.

Steps

1. Setup

Install required libraries:

pip install pandas numpy statsmodels matplotlib

2. Data

The program uses randomly generated price data for two assets.

3. Cointegration Test

Use coint() to check if the assets are cointegrated:

• If the p-value < 0.05, the assets are cointegrated and suitable for Stat Arb.

4. Spread Modeling

A linear regression predicts one asset’s price using the other, and the spread is calculated as the difference between actual and predicted prices.

5. Trading Signals

• Entry Threshold: Spread > (Mean ± 2 * Std Dev).
• Exit Threshold: Spread reverts to the mean.

6. Backtesting

Simulates trades based on the signals, calculates portfolio value, and tracks performance over time.

Example Output

Cointegration Test p-value: 0.3424181037499916
Spread Mean: -5.3660187404602763e-14
Spread Std Dev: 5.106547769214354
Final Portfolio Value: 99777.6834577597

The script will also generate the following plots:

1. Portfolio Value Over Time: A graph showing how the portfolio value evolves over time.
2. Spread with Entry and Exit Thresholds: A plot displaying the spread between the two assets, along with entry and exit thresholds and marked entry/exit points.

Customization

• Data Input: Replace the simulated data with actual price data for the assets you want to trade.
• Entry/Exit Thresholds: Adjust the entry and exit thresholds based on your desired risk/reward profile.
• Initial Capital: Change the initial_cash variable to reflect your starting portfolio value.

Notes

This is a basic implementation to learn the Statistical Arbitrage Strategy. To improve:

• Use real data (e.g., stock prices).
• Account for transaction costs and risk.
• Experiment with different thresholds.

Run the program, visualize the results, and tweak to deepen your understanding!

Statistical Arbitrage: Coca-Cola vs PepsiCo

updated script: statistical-arbitage-pep-ko.py

This script implements a statistical arbitrage strategy using historical stock data for Coca-Cola (KO) and PepsiCo (PEP). It performs the following steps:

• Data Collection: Fetches historical stock prices from Yahoo Finance (2015-2020).
• Cointegration Test: Checks if the assets are cointegrated with a p-value threshold of 0.1.
• Spread Modeling: Calculates the spread between the two stock prices using linear regression.
• Entry/Exit Signals: Generates signals based on spread deviations from the mean.
• Backtesting: Simulates trades and tracks portfolio performance.
• Results: Outputs final portfolio value and plots performance.

Dependencies:

pip install yfinance statsmodels matplotlib pandas numpy

License

MIT License